Spaces:
Sleeping
Sleeping
File size: 7,477 Bytes
4a695e8 a7743a4 f63e910 57f316a f63e910 57f316a f63e910 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 | import re
from src.pdf_parser import extract_text_from_pdf
from src.grobid import extract_metadata_grobid
def extract_title(text):
"""Extracts paper title (before Abstract), avoiding citations and irrelevant headers."""
lines = [l.strip() for l in text.split("\n") if l.strip()]
# Find index of "Abstract" (case-insensitive) or common translations
abs_index = None
for i, line in enumerate(lines):
low = line.lower()
if low.startswith("abstract") or any(k in low for k in ["résumé", "resumen", "summary", "overview", "abstract—"]):
abs_index = i
break
# If Abstract found, check lines before abstract; else check top area
if abs_index is not None:
candidate_lines = lines[:abs_index]
else:
candidate_lines = lines[:100]
skip_words = ["journal", "doi", "copyright", "arxiv", "volume",
"methods", "open access", "citation", "editor", "published"]
candidates = []
for line in candidate_lines:
low = line.lower()
# Skip unwanted lines
if any(w in low for w in skip_words):
continue
# Skip author-like lists (many commas) or emails
if "@" in line or re.search(r"\bjournal\b", low):
continue
# Skip short/very long lines
if 5 <= len(line.split()) <= 30:
# avoid lines that look like "Editor: Name" etc
if re.match(r"^(editor|edited by|edited).+", low):
continue
candidates.append(line)
# Prefer line with colon (typical title: subtitle)
for line in candidates:
if ":" in line:
return line
# fallback: longest candidate
if candidates:
return max(candidates, key=len)
return "Untitled"
def extract_authors(text, max_scan_lines=80):
"""
Flexible author extraction:
- Finds a likely title, then scans the following lines looking for names.
- Handles comma-separated authors on one line or names on separate lines.
"""
lines = [ln.strip() for ln in text.splitlines() if ln.strip()]
authors = []
# 1) find a likely title index (first reasonably long line)
title_idx = None
for i, line in enumerate(lines[:40]):
if len(line.split()) >= 4 and not any(k in line.lower() for k in ["journal", "doi", "copyright", "arxiv"]):
title_idx = i
break
start = title_idx + 1 if title_idx is not None else 0
# 2) scan lines after title until we hit abstract/keywords or a limit
abstract_markers = ["abstract", "résumé", "resumen", "summary", "overview", "introduction", "keywords"]
for line in lines[start:start + max_scan_lines]:
low = line.lower()
if any(m in low for m in abstract_markers):
break
# If line contains commas or " and ", likely multiple authors
if "," in line or " and " in low or ";" in line:
parts = re.split(r",|;|\sand\s", line)
for part in parts:
part = part.strip()
if not part:
continue
# match typical name patterns: "First Last", "First M. Last", "Last, First"
# handle "Last, First" by swapping
if re.match(r"^[A-Z][a-z]+,\s*[A-Z][a-z]+", part):
# swap "Last, First" -> "First Last"
t = re.split(r",\s*", part)
name = t[1] + " " + t[0]
else:
name = part
# require at least two capitalized tokens to be considered a name
cap_tokens = [t for t in name.split() if re.match(r"^[A-Z][a-z]+\.?$", t) or re.match(r"^[A-Z]\.$", t)]
if len(cap_tokens) >= 2 and not looks_like_affiliation(name):
authors.append(name)
if authors:
continue
else:
# attempt single-line name extraction
matches = re.findall(r'\b[A-Z][a-z]+(?:\s[A-Z]\.?\s?[A-Z][a-z]+){0,2}\b', line)
for m in matches:
if len(m.split()) >= 2 and not looks_like_affiliation(m):
authors.append(m)
# 3) fallback: check metadata-like "By X Y" lines
if not authors:
for line in lines[:40]:
m = re.match(r"^(By|BY|by)\s+(.+)$", line)
if m:
parts = re.split(r",|;|\sand\s", m.group(2))
for p in parts:
p = p.strip()
if p and not looks_like_affiliation(p):
authors.append(p)
if authors:
break
return authors if authors else ["Unknown"]
AFFIL_KEYWORDS = [
"university", "institute", "department", "school", "hospital",
"clinic", "center", "centre", "laboratory", "lab", "college",
"medicine", "research", "faculty", "division", "program", "department of"
]
def looks_like_affiliation(line):
low = line.lower()
if any(k in low for k in AFFIL_KEYWORDS):
return True
if "@" in line or "http" in low or "www." in low:
return True
# if line has few capitalized tokens relative to total tokens, it's likely not a name
toks = [t for t in line.split() if t.strip()]
if len(toks) == 0:
return True
cap = sum(1 for t in toks if re.match(r"^[A-Z][a-z]+$", t))
if cap / len(toks) < 0.4:
return True
return False
def extract_abstract(text):
"""Extract abstract block (multiple formats supported). Gathers until next main heading."""
lines = text.split("\n")
abstract_lines = []
capture = False
abstract_markers = ["abstract", "résumé", "resumen", "summary", "overview"]
stop_markers = ["introduction", "keywords", "1.", "methods", "materials", "results", "conclusion", "references"]
for line in lines:
l = line.lower().strip()
if any(m in l for m in abstract_markers):
capture = True
continue
if capture:
# stop when a common section heading appears
if any(re.match(rf"^{sm}\b", l) for sm in stop_markers):
break
abstract_lines.append(line.strip())
abstract = " ".join(abstract_lines).strip()
return abstract
def extract_metadata(pdf_file):
"""
Extract metadata using GROBID first.
If GROBID fails, use the regex-based fallback.
"""
# Read full paper text once
text = extract_text_from_pdf(pdf_file)
# Reset pointer because extract_text_from_pdf() consumed it
pdf_file.seek(0)
# -------------------------------
# Try GROBID
# -------------------------------
try:
grobid_meta = extract_metadata_grobid(pdf_file)
if grobid_meta:
title = grobid_meta.get("title", "").strip()
authors = grobid_meta.get("authors", [])
abstract = grobid_meta.get("abstract", "").strip()
return {
"title": title if title else extract_title(text),
"authors": authors if authors else extract_authors(text),
"abstract": abstract if abstract else extract_abstract(text)
}
except Exception:
pass
# -------------------------------
# Fallback (Regex)
# -------------------------------
return {
"title": extract_title(text),
"authors": extract_authors(text),
"abstract": extract_abstract(text)
}
|